KDE on KDE 4.0 (Groklaw) [LWN.net]

KDE on KDE 4.0 (Groklaw)

Posted Jul 12, 2008 17:17 UTC (Sat) by rsidd (subscriber, #2582) [Link]

I think "rare, obscure, corner-case issues" exist in all software of moderate complexity
(except maybe TeX) and are not a reason for holding back a release.  The thing people (me
included) didn't like about the KDE 4.0 release was that the issues were not rare or
corner-case -- they were glaringly obvious on a default install.  And it wasn't just missing
features -- it was crashes (particularly of plasma), random misdrawings of the screen, and so
on.  

I'm now running KDE 4.1beta2 on one computer and am reasonably impressed with what they have
achieved in the past six months.  It is much less buggy and much more feature-complete, but
there are still huge missing items (especially to do with configurability -- you can't even
move icons on the panel, for example).  I think KDE 4.1 would have been a fair "public beta"
candidate, and if one wanted to stretch a point, it could even have been the 4.0 release.  But
the 4.0 that was actually released, even if advertised loudly as "developers and adventurous
users only", was a bit of a shortchange.

KDE on KDE 4.0 (Groklaw)

Posted Jul 12, 2008 18:52 UTC (Sat) by Janne (guest, #40891) [Link]

"It is much less buggy and much more feature-complete, but
there are still huge missing items (especially to do with configurability -- you can't
evenmove icons on the panel, for example)."

Then you are happy to hear that they implemented the moving of icons 'n stuff in the panel
after beta2 was released. So it will work in 4.1

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 18:18 UTC (Sun) by sebas (subscriber, #51660) [Link]

As we said in the article, it was a basic finished desktop, but not for everyone. I, for
example did use KDE 4.0 for all my daily work. That makes it at least suitable for some (which
is what we claimed).

By the way, there is no promise to make KDE 4 the same as 3.5, we'll make it better, whatever
that means, but certainly different. You're welcome to hack on your pet feature though. We're
hiring ;-)

KDE on KDE 4.0 (Groklaw)

Posted Jul 14, 2008 6:31 UTC (Mon) by Cato (subscriber, #7643) [Link]

Since it was "not for everyone", why wasn't it labelled as a Beta, Technology Preview, or
similar?  I use KDE as my desktop environment, and tried KDE 4.0 very soon after release via
Live CD, without realising quite how incomplete it was.  As a result, I wrote a number of
comments on the feedback Wiki page saying how much work should have been done before it was
released as plain "4.0" - this episode has not improved my confidence in the KDE project, to
say the least.

It would really help matters if the KDE developers just admitted they made a big mistake on
the labelling of this release - the content was fine but it was a technology preview not a
finished, usable release.

KDE on KDE 4.0 (Groklaw)

Posted Jul 12, 2008 17:22 UTC (Sat) by boudewijn (subscriber, #14185) [Link]

If it doesn't make sense to you it's probably because you're not a 
developer, so you wouldn't be aware how these things work. To start with 
your second point: our experience is that nobody installs and tests 
alphas, betas or release candidates anymore. Which means no user 
testing -- basically, the users are reneging on their responsibility. 
(Not all of them, we've got Christophe Giboudeaux who's testing our Krita 
alpha releases and he's absolutely a brilliant bug reporter. But there 
aren't many like him.)

But that's not the most important thing. Almost all of KDE 4.0 was pretty 
stable, except for the the desktop component.

The reason plasma wasn't stable is directly related to your first 
question. Until there's are release, libraries and api's aren't stable. 
It's a simple matter of fact, and you can either understand it or not, 
but it doesn't change the fact that right up to the first rc, developers 
will break binary and source compatibility. 

Even more importantly, there are many app developers who are _not_ 
comfortable with tracking subversion or alpha and beta releases. They 
want to install a final release of the libraries and code against that. 
And the libraries in KDE 4.0 were absolutely fine. We're still targetting 
KDE 4.0 with KOffice, for instance.

KDE 4.0 was ready for developers and adventurous users. KDE 4.0.4 is 
ready for daily use, if you use your computer for work and not for 
tweaking your desktop. KDE 4.1rc1 is a stable, featureful and great 
dekstop.

KDE on KDE 4.0 (Groklaw)

Posted Jul 12, 2008 22:49 UTC (Sat) by mikov (subscriber, #33179) [Link]

If it doesn't make sense to you it's probably because you're not a developer, so you wouldn't be aware how these things work.

I've been a developer for more than 15 years, thank you, so I know a thing or two about how these things work :-) But that's not the point - I will be the first to admit that managing and releasing a free software project is very different from a proprietary one, and I only have major first-hand experience with the latter. I think I understand and strongly sympathize with the KDE developers, but I still think that ultimately they made a mistake.

To start with your second point: our experience is that nobody installs and tests alphas, betas or release candidates anymore. Which means no user testing -- basically, the users are reneging on their responsibility.

I hear you about users not wanting to test beta versions any more. This is a serious problem, but I suspect that shoving beta software to unsuspecting users may not be the best approach. BTW, the distributions who bundled KDE 4.0 by default are also very much to blame here.

KDE is a victim of its own success - we have gotten used to very polished releases, with practically no visible bugs, and every new release adding more flexibility and functionality without removing any. KDE4 broke that tradition, and whatever the justifications are, the backslash is undeniable. Even Vista is more backward compatible with XP than KDE4 with KDE3 !

Take me for example. As far as KDE is concerned I am strictly a user. I use my desktop for work and I am so used to it, and so productive with it (again, thank you KDE!), that I don't even notice it anymore. This is how it should be. If I upgrade to a new version of KDE, I need everything to remain exactly the same, so I can immediately continue to be productive, and I can explore the exciting new features at my own pace, when and if I have time.

I think that this is unavoidable as Linux (and KDE and Gnome, etc) become more widely accepted and established. The projects just can't afford to make such drastic changes.

Fortunately, I don't think what happened with KDE4 is such a big deal and I am sure the backslash will subside eventually. I am reasonably hopeful that Debian (my platform of choice) will not make a stable release with KDE4 before it is fully ready.

KDE on KDE 4.0 (Groklaw)

Posted Jul 17, 2008 21:29 UTC (Thu) by lysse (guest, #3190) [Link]

If it doesn't make sense to you it's probably because you're not a developer
I've been a developer for more than 15 years, thank you

Today's mantra: "Any assumption I make about someone querying me is certain to bite me in the ass when they read my reply". I've seen so many exchanges like the above, I can't believe I'm still seeing them...

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 8:17 UTC (Sun) by rsidd (subscriber, #2582) [Link]

Was there much user testing of KDE 4.0?  Speaking for myself, I tried it, couldn't believe how
unfinished it was, and left.  I didn't bother reporting any bugs -- the issues were glaringly
obvious, and aseigo made it clear that he knows what all the issues are and either they're
being worked on or the user is misguided for asking.  I don't think I'd have been alone.  It
speaks a lot of the goodwill the KDE project has that many people (including me) are now
willing to give 4.1 a try, and it speaks well of what has been achieved that we are now
willing to ignore its rough edges (of which many remain).  I don't think anything would have
been lost by delaying the release by another 6 months and calling the current version 4.0.
But that's all history now; if things go well I don't think there will be a long-term impact.

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 18:23 UTC (Sun) by sebas (subscriber, #51660) [Link]

Application developers need their apps released, that's not possible without the libs
released. Holding back the release of stable libs drives app developers away for they don't
have a perspective to get their software released. They're relying on us here.

Then, it's not only our own (KDE developers, KDE app developers) but also other pieces in the
stack. We've run into grave bugs in Qt, X, GTK+, various video drivers among which intel and
nvidia, xine and other components. Some of those need released software to even justify
testing with new KDE. Some of those bugs still exist (most notably nvidia performance
problems). We're pushing not only the boundaries of our own code, but also that of others, and
those also need some time to catch up.

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 18:58 UTC (Sun) by mikov (subscriber, #33179) [Link]

I think it is clear that this is not a problem with KDE4 itself, but with how it was
communicated to the public. Regardless of the technical motivations for releasing, there is no
denying that even KDE 4.1 is not yet ready to be used as a replacement for KDE3 by ordinary
users - not by hackers or KDE developers, but people who use their PC for office tasks, etc.

Granted, it is primarily the distribution's responsibility to decide whether to bundle KDE4 as
default or not. As far as I know, OpenSUSE and especially Fedora are not aimed at "ordinary"
users, so the complaints by people who downloaded the distro to "try KDE 4", but don't
actually use it for their day-to-day job, are greatly exaggerated and borderline dishonest !

Regardless, I think that you guys should take the high road and apologize for the
miscommunication and make it even more clear that KDE 4 is not yet ready for mass adoption. In
retrospect, I think that adding a suffix to the release name (e.g. "-DR1" or whatever) would
have been ideal. I am sure that such an announcement would be taken very positively by
everybody and would silence the majority of the complaints.

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 21:05 UTC (Sun) by sebas (subscriber, #51660) [Link]

If we would've waited until aunt tillie can use our desktop, then we would release some months
after the last developer has left the project.

I think that expecting a release not before KDE 4 is usable for everyone is just not
realistic, moreover, it's also not how Free Software development works. In fact, we were
continuously asking ourselves the question when KDE 4 would be good enough for a first
release, being aware that we probably cannot get it perfect the first time, especially not WRT
to Plasma being brand new and all.

Frankly, for the part I was involved with, I'd do the release scheduling again like this, just
try to communicate more clearly who it's meant for. The communication around has not been
perfect, although we tried to make it clear enough.

As I already said in the other thread, releasing is not just for the users, it also plays an
essential role for the community itself -- and that's what we focussed on. We simply put the
health of the project above the gusto of some users that -- in retrospect -- didn't even
follow closely enough the purpose and status of this release.

As opposed to saying "We were wrong, we shouldn't have released 4.0 at that stage", I'd say
"Next time, we'll try to make it more clear what our target audience for a release is".

I hope this makes it easier to understand the constraint leading up to the release of KDE 4.0.

KDE on KDE 4.0 (Groklaw)

Posted Jul 13, 2008 21:22 UTC (Sun) by mikov (subscriber, #33179) [Link]

Well said.

I just want to make sure that one important point is not lost in the noise:

KDE is a great desktop. Some trouble is unavoidable from time to time, but you guys are doing
an excellent job. I and the overwhelming majority of users are very grateful for your work and
dedication. Keep up the good work!

Don't let the (hopefully constructive) criticism bother you - we, the users, are not
ungrateful, just spoiled :-)

KDE on KDE 4.0 (Groklaw)

Posted Jul 19, 2008 10:15 UTC (Sat) by malor (subscriber, #2973) [Link]

We simply put the health of the project above the gusto of some users

And that's exactly the problem. You chose to screw over the early adoption crowd by releasing software that was glaringly broken, and calling it 'stable'. Point-zero releases mean something to the rest of the world, and blaming the users for not understanding that YOUR use of that release term was different than everyone else's is entirely disingenuous.

YOU blew it, not the users. You cared more about some nebulous project goals than the actual people who are trying to get real work done with it. And you didn't even get what you wanted; any short-term bug reports you may have gotten have been entirely outweighed by the credibility loss you've suffered. You're further compounding that loss by now blaming users for your mistake.

By losing credibility, you lose users, which means you also lose testers. So, to get a short-term testing bump, you've sacrificed future testing. You ate a bunch of your seed corn.

If you want more testers, you need to recruit them honestly, not by deception. You recruit testers from the pool of users, and you get users by providing stable software releases that are actually stable.

If you try to fool users into being testers, they'll stop being both.

KDE on KDE 4.0 (Groklaw)

Posted Jul 19, 2008 10:47 UTC (Sat) by nix (subscriber, #2304) [Link]

Yeah. Point-zero releases mean 'unstable and dangerous, early adopters 
only' to the rest of the world, and have for *decades*.

KDE on KDE 4.0 (Groklaw)

Posted Jul 19, 2008 14:38 UTC (Sat) by malor (subscriber, #2973) [Link]

A point-zero release means that the devs think a piece of software is stable and feature-complete. It's a claim that it's done and ready for public consumption. Now, reality often differs from that. Some organizations do better than others, but at least the devs are trying. If a point-zero release is broken, it's at least inadvertent.

In this case, neither of those things were true. It was a point zero release that the devs knew perfectly well was very unstable, with an API still in flux. Not everything was even written yet. Calling that a .0 release, no matter how many caveats you hang off it, is guaranteed to mislead users who aren't really paying attention that day, or who figure that the people warning them away are overreacting; this is KDE, after all. How badly could it be broken?

Call it 4.0 beta something and everyone is happy. Nobody gets mad if it falls apart into smoking pieces. Beg and plead for testers if needed. Calling it even an RC candidate would have been misleading, since it wasn't ready to be released yet.

Giving it full point-zero status was an outright lie. It netted them a minor short-term gain with a probable larger long-term loss; how many users are on GNOME now because of this? Even if it's just a few hundred, that's a real loss for a gain that appears, as far as I can see, entirely theoretical.

KDE on KDE 4.0 (Groklaw)

Posted Jul 19, 2008 14:41 UTC (Sat) by malor (subscriber, #2973) [Link]

(oh, and before you start listing commercial software with borked .0 releases; payware
sometimes has to ship to make the quarter.  Open source has no such need.  There's no fiscal
pressure to release before a product is ready.)

KDE on KDE 4.0 (Groklaw)

Posted Jul 20, 2008 19:08 UTC (Sun) by mikov (subscriber, #33179) [Link]

Technically, I think you are right. KDE lost credibility and potential users/testers because
of this. 

But I also think that there is no need for all this negativity. Everybody is bound to screw up
once in a while. While they failed in labeling the  release or communicating it, they didn't
fail as developers - the technology is great (or will be) - and that's what really matters. 

I think it is important for KDE users and supporters not to make a too big deal out of it.